Location-free image audio reversible data hiding method based on additive vector superposition

Abstract

The invention relates to the technical field of data hiding, and particularly provides a positioning-free graph audio reversible data hiding method based on additive vector superposition. The method comprises the following steps: carrying secret data by adopting a specific orthogonal extension basis vector based on an additive vector superposition data hiding algorithm, and superposing the secret data into an object audio stream; designing a positioning-free graph scheme by comparing the correlation between the inner product of the base vector and the content vector and the base vector 2-norm; according to the method, the adaptive predictor is designed, the value of the target audio point is estimated with high precision through the adjacent samples according to the prediction precision requirement, the distortion of the data hiding audio stream is reduced, and the data hiding capability of the audio stream is improved.

Description

Technical Field

[0001] This invention relates to the field of data hiding technology, and in particular to a localization-free, reversible audio data hiding method based on additive vector superposition. Background Technology

[0002] With the development of social media networks, an increasing amount of clandestine information is being transmitted through secure communication channels. Data hiding technology enables functions such as digital steganography, clandestine communication, and copyright protection by hiding secret data within host signals. As an important branch of data hiding technology, reversible data hiding (RDH) can recover hidden secret data and host signals without losing any information. Therefore, RDH has significant application potential in multiple fields such as military, medical, copyright authentication, and covert communication. Typically, the performance of an RDH scheme is evaluated based on two key indicators: data hiding capability and host signal distortion. A good RDH algorithm can maintain minimal host signal distortion even when processing large amounts of secret data.

[0003] The initial approach of RDH utilized lossless compression algorithms to create space for hiding secret data. First, the image was compressed, and data was hidden in the free space, achieving reversible data hiding. Further, the least significant bit (LSB) of the image was losslessly compressed, creating free space for image hashing hiding. A novel RDH method, Differential Expansion (DE), was proposed, which hides data by expanding the difference between adjacent pixels. As long as the modified pixel value does not overflow or underflow, the data can be hidden in the least significant bit of two expandable pixels. This method significantly improves the data hiding capability of RDH schemes to a certain extent, surpassing compression-based schemes. Subsequently, an RDH scheme based on the Image Histogram Shift (HS) algorithm was proposed, which hides secret data by adjusting the peak and valley pixels of the image histogram. However, the data hiding ability of natural images is limited because their histograms are usually relatively flat. To further improve this, a sparse prediction error plane was constructed to achieve a steep distribution of the image histogram. High-capacity data hiding is achieved by moving the prediction error most frequently. Since the prediction error histogram always exhibits a Laplace distribution, the data hiding ability is effectively improved.

[0004] In audio RDH schemes, secret data is typically processed and reversibly hidden into the amplitude of the audio stream, with the goal of keeping audio stream distortion to a very low level even after a large amount of data is hidden. Although RDH technology has made significant progress in recent years, error prediction and data hiding algorithms have become increasingly complex. Simultaneously, the perceptual quality of the audio stream deteriorates rapidly with increasing data hiding load. Summary of the Invention

[0005] In view of this, the present invention provides a localization-free audio reversible data hiding method based on additive vector superposition, which reduces the distortion of the carrier audio stream after data hiding and improves the data hiding capability of the audio stream.

[0006] In a first aspect, the present invention provides a location-free, graph-based, reversible audio data hiding method based on additive vector superposition, the method comprising: Step 1: Based on the additive vector superposition data hiding algorithm, a specific orthogonal extended basis vector is used to carry secret data and hide it in the object audio stream; Step 2: Based on Step 1, design a location-free graph solution by comparing the inner product of the basis vectors and the content vectors with the 2-norm of the basis vectors. Step 3: Based on Step 2, design a predictor that adaptively estimates the value of the target audio point through its neighboring samples, generates a sparse error prediction matrix, and hides the data based on the sparse error prediction matrix to improve the capacity of reversible data hiding.

[0007] Optionally, step 1 includes: A. Data hiding: Assume the secret data sequence is , Secret position Convert to watermark data bits using equation (1) : (1) Based on equation (1), the secret data sequence is converted to: , ; Design basis vectors , Its elements take values ​​of 1 and -1, and its length is l The sum of all elements is zero; construct the set of basis vectors. Different basis vectors are orthogonal to each other; Assumption A It is a length of N The audio stream, from A Selecting adjacent samples to form a content vector Its expression is: , Let the length of the content vector also be... l The length of the content vector set is consistent with that of the basis vectors; the content vector set is represented as... ; According to the additive vector superposition data hiding principle, when the vectors satisfy the condition ,and At that time, the secret information bits are hidden in the content vector according to formula (2). middle; (2) , (3) in, , , which indicates the first t A content vector that hides secret data; It is the secret bit to be hidden, with a value of 1 or -1; They are mutually orthogonal basis vectors; It is the number of hidden secret bits; on the other hand, when the vector satisfies the condition ,and At that time, according to formula (2), it will make or The basis vectors that hold true are superimposed on the content vector. Finally, the data hiding matrix is ​​obtained. ; When the amount of secret data is large, the orthogonality of the basis vectors can be used to hide the data sequence. The data is segmented and repeatedly superimposed onto the content vector to increase the data hiding capacity of the carrier audio stream, and its expression is given by formulas (4)-(6): (4) (5) (6) in, It is the set of selected orthogonal basis vectors; This represents the audio data matrix after multi-level data hiding; It is the number of times the data was hidden. It is the data hiding strength coefficient; B. Data Extraction: Assumption Matrix It can be seen that, composed of the data hidden content vectors in the received audio stream, , ,when Then, the hidden data can be extracted without loss according to formula (7): (7) No. t The basic principle of extracting the secret bits is shown in formula (8): (8) because and The values ​​of are all positive integers, therefore, when the condition is met... , At that time, the hidden data is extracted without loss according to formula (9): (9) When the condition is met , At that time, according to the formula The hidden data is extracted and discarded without loss, thus enabling the receiving end to recover the original secret information from the audio stream that carries the hidden data without loss. C. Original audio restoration: The original content vector is extracted from the data-hidden carrier audio stream, and the original audio stream can be reconstructed by subtracting the superimposed weighted basis vectors, as shown in formula (10): (10) Optionally, step 2 includes: According to formula (9), when the condition is satisfied... and When, a bit is reversibly hidden into the carrier vector; when At this point, the inner product of the content vector and the basis vector is positive. Then, by additively superimposing the hidden data 1, we obtain... The receiving end can use this to determine that the vector has not been signed into secret data; when At this point, the inner product of the content vector and the basis vector is negative. Then, the hidden data -1 is added additively, resulting in... Based on this, the receiving end determines that the vector also does not hide secret data, thus achieving reversible information hiding of audio data without using a positioning map.

[0008] Optionally, step 3 includes: According to the accuracy requirements of the hidden data, high-precision target point prediction can be achieved by adaptively adjusting the number of audio points involved in the prediction; large-capacity data hiding can be achieved by using the sparse error plane formed by the error of accurate prediction, as shown in formula (11): (11) in, , This is the prediction error value. It is the target point. yes The predicted value, , , and , , These are the nearest neighbor samples before and after the predicted target audio point. s It is the distance between the sample point and the target audio point; when s When we take 3, we get: (12) I. The process by which the sending end hides secret data in the audio stream: a. Divide the audio stream into point sets and intersection sets; using formula (11), through its 2 n The object audio point values ​​in the intersection set of the neighboring points are predicted to construct a sparse prediction error matrix. b. Divide the secret data into two equal parts, and according to step cd, hide the first part in the intersection set; c. Divide the cross set into two parts: the data hiding part E and the reserved part R; extract the least significant bit (LSB) of each audio point in the reserved part, and save the auxiliary data into the free LSB; d. Based on the location-free map scheme, the key data and least significant bit of the audio points are hidden in the reserved part and combined with part E; e. Using formula (12), predict the object audio points in the point set and their 2 n The values ​​of the nearest neighbor points are used to construct a sparse prediction error matrix for the invertible data hiding RDH; f. Use the same cross-set strategy as in step cd to hide the secret data in the point set; g. Construct a data-hidden audio stream using the fork set and point set data after hiding the information; II. The process of recovering the secret data and the original audio stream at the receiving end, and extracting the secret data: h. Divide the received audio stream into an intersecting set and a dotted set; i. Extract auxiliary information from the LSB of the retained part of the point set, and construct the prediction error matrix of the data hidden part E in the point set using formula (11); j. Based on the content of the extracted auxiliary data, extract the secret data and the LSB of the reserved part from the prediction error matrix of the data hidden part E in the point set based on the location-free map scheme; k. Recover the original amplitude of each audio point in the hidden data portion E and the point set preserved portion R; l. Using the same point set strategy as in steps j-k, extract the hidden secret data from the cross set and then restore the original value of each audio point; m. Reconstruct secret data using information extracted from point sets and intersection sets, and recover the original audio stream using the recovered audio points.

[0009] In a second aspect, embodiments of the present invention provide a computer-readable storage medium comprising a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to execute the localization-free audio reversible data hiding method based on additive vector superposition in the first aspect or any possible implementation thereof.

[0010] Thirdly, embodiments of the present invention provide an electronic device, including: one or more processors; a memory; and one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions that, when executed by the device, cause the device to perform the localization-free audio reversible data hiding method based on additive vector superposition in the first aspect or any possible implementation of the first aspect.

[0011] The technical solution provided by this invention includes a data hiding algorithm based on additive vector superposition, which uses specific orthogonal extended basis vectors to carry secret data and hides it in the target audio stream; a location-free map scheme is designed by comparing the inner product of the basis vector and the content vector with the 2-norm of the basis vector; and a predictor is designed to estimate the target audio point through its neighboring samples. This method reduces audio distortion caused by data hiding and improves the data hiding capability of the audio stream. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart of a localization-free audio reversible data hiding method based on additive vector superposition provided in an embodiment of the present invention; Figure 2 A flowchart for hiding secret data at the sending end provided in an embodiment of the present invention; Figure 3 A flowchart for extracting secret data at the receiving end provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the signal-to-noise ratio of data-hidden audio streams with and without localization maps at a load rate of 0.5 bpp, provided by an embodiment of the present invention. Figure 5 This is a schematic diagram illustrating the signal-to-noise ratio of data-hidden audio streams with and without localization maps at a load rate of 1.0 bpp, provided by an embodiment of the present invention. Figure 6This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] It should be understood that the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0016] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in the embodiments of this invention are also intended to include the plural forms unless the context clearly indicates otherwise.

[0017] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0018] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0019] Figure 1 The flowchart of the localization-free audio reversible data hiding method based on additive vector superposition provided in the embodiments of the present invention is as follows: Figure 1 As shown, the method includes: Step 1: Based on the additive vector superposition data hiding algorithm, a specific orthogonal extended basis vector is used to carry secret data and hide it in the object audio stream.

[0020] Audio streams are one-dimensional dynamic signals that change over time. Due to the inherent correlation between adjacent samples, secret data can be reversibly hidden into the audio stream using an RDH scheme based on additive vector superposition data hiding. Additive vector superposition data hiding is a spectrum spreading technique commonly used for secure data transmission in wireless communications. Communication systems based on additive vector superposition transmit secret signals by using different orthogonal basis vectors; these signals can only be correctly decoded using the same basis vectors. Therefore, multidimensional data can be transmitted within a single channel without interference, thus improving the system's transmission capacity and security. The RDH framework based on additive vector superposition data hiding can be viewed as a covert communication system where the host audio stream serves as a common transmission channel, and the secret data represents the signal to be transmitted. In an RDH system based on additive vector superposition data hiding, the elements of the orthogonal spread basis vectors consist of 1 and... The vectors are composed of equal amounts, and the sum of the vector elements is kept to be 0. Due to the orthogonality and zero mean of the basis vectors used, the secret data can be repeatedly hidden into the target audio stream. In the process of multi-level data hiding, most of the elements of the extended vectors will cancel each other out, thus enabling the audio RDH scheme based on additive vector superposition data hiding to achieve efficient data hiding.

[0021] In this embodiment of the invention, step 1 includes: A. Data hiding: Assume the secret data sequence is , Secret position Convert to watermark data bits using equation (1) : (1) Based on equation (1), the secret data sequence is converted to: ., ; Design basis vectors , Its elements take values ​​of 1 and -1, and its length is l The sum of all elements is zero; construct the set of basis vectors. Different basis vectors are orthogonal to each other; Assumption A It is a length of N The audio stream, from A Selecting adjacent samples to form a content vector Its expression is: , Let the length of the content vector also be l The length of the content vector set is consistent with that of the basis vectors; the content vector set is represented as... ; According to the additive vector superposition data hiding principle, when the vectors satisfy the condition ,and At that time, the secret information bits are hidden in the content vector according to formula (2). middle; (2) (3) in, , Indicates the first t A content vector that hides secret data; It is the secret bit to be hidden, with a value of 1 or -1; They are mutually orthogonal basis vectors; It is the number of hidden secret bits; on the other hand, when the vector satisfies the condition ,and At that time, according to formula (2), it will make or The basis vectors that hold true are superimposed on the content vector. Finally, the data hiding matrix is ​​obtained. ; When the amount of secret data is relatively large (for example, there is...) n=m·k (bits), the orthogonality of basis vectors can be used to hide the data sequence to be hidden. The data is segmented and repeatedly superimposed onto the content vector to increase the data hiding capacity of the carrier audio stream, and its expression is given by formulas (4)-(6): (4) (5) (6) in, It is the set of selected orthogonal basis vectors; This represents the audio data matrix after multi-level data hiding; It is the number of times the data was hidden. It is the data hiding strength coefficient; B. Data Extraction: Assumption Matrix It can be seen that, composed of the data hidden content vectors in the received audio stream, , ,when When this happens, the hidden data can be extracted without loss of quality using formula (7): (7) No. t The basic principle of extracting the secret bits is shown in formula (8): (8) because and The values ​​of are all positive integers, therefore, when the condition is met... , At that time, the hidden data is extracted without loss according to formula (9): (9) When the condition is met , At that time, according to the formula The hidden data is extracted and discarded without loss, thus enabling the receiving end to recover the original secret information from the audio stream that carries the hidden data without loss. C. Original audio restoration: The original content vector is extracted from the data-hidden carrier audio stream, and the original audio stream can be reconstructed by subtracting the superimposed weighted basis vectors, as shown in formula (10): (10) By leveraging the orthogonality of basis vectors to achieve multi-level data hiding, the reversible data hiding performance of audio streams is significantly improved. Furthermore, since these elements consist of 1 or... The proposed scheme consists of 1 basis vector, most of which cancel each other out during the multi-level data hiding process; this ensures that the audio stream remains perceptually high even with large-volume data hiding. Furthermore, only the receiver, possessing the same basis vectors as the sender, can fully recover the hidden secret data and the original audio stream; this further enhances the security of the proposed scheme.

[0022] Step 2: Based on Step 1, design a location-free graph solution by comparing the inner product of the basis vectors and the content vectors with the 2-norm of the basis vectors.

[0023] Based on the above discussion, some carrier vectors may not be suitable for data hiding. Therefore, identifying the audio points corresponding to these carrier vectors is crucial for achieving reversible data hiding. Traditionally, location maps are used to indicate audio points that are unsuitable for data hiding. However, adding location maps significantly reduces the ability to hide secret data. Therefore, a location map-free scheme is designed to achieve higher data hiding capabilities.

[0024] In this embodiment of the invention, step 2 includes: According to formula (9), when the condition is satisfied... and When, a bit can be reversibly hidden into the carrier vector; when At this point, the inner product of the content vector and the basis vector is positive. Then, by additively superimposing the hidden data 1, we obtain... The receiving end can use this to determine that the vector has not been signed into secret data; when At this point, the inner product of the content vector and the basis vector is negative. Then, the hidden data -1 is added additively, resulting in... The receiving end can then determine that the vector does not hide secret data, thus achieving reversible information hiding of audio data without using a positioning map.

[0025] Therefore, with this scheme, both the hidden secret bits and the original audio stream can be completely recovered using the location-free method; this scheme can achieve reversible data hiding without a location map, thus significantly saving storage space for the secret data.

[0026] Step 3: Based on Step 2, design a predictor to estimate the target audio point using its neighboring samples.

[0027] Based on the above analysis, when the carrier vector meets certain conditions, secret data can be hidden in the carrier audio stream. This is because 1 and 2 in the extended vector... Since the number of 1s is the same, the calculated value is equal to the sum of the differences between adjacent elements. Therefore, the more similar the adjacent samples are, the smaller the calculated value, thus minimizing the distortion caused by data hiding while hiding more data.

[0028] In this embodiment of the invention, step 3 includes: Based on the close correlation between adjacent samples, a content vector is constructed using the prediction error of the audio signal amplitude for data hiding. This makes the elements in the content vector more similar, thereby improving the data hiding performance of the audio RDH scheme. Considering that most adjacent audio samples have significant similarity, which is largely influenced by their positions, the prediction error is determined by calculating the average of the samples before and after the object point, as shown in formula (11): According to the accuracy requirements of the hidden data, high-precision target point prediction can be achieved by adaptively adjusting the number of audio points involved in the prediction. Large-capacity data hiding can be achieved through the sparse error plane formed by the error of accurate prediction. It is shown in formula (11): (11) in, , This is the prediction error value. It is the target point. yes The predicted value, , , and , , These are the nearest neighbor samples before and after the predicted target audio point. s It is the distance between the sample point and the target audio point; In particular, whens When we take 3, we get: (12) Due to the close correlation between adjacent audio samples, the proposed scheme obtains a small prediction error value, which is concentrated around 0, as shown in Table 2.

[0029] Table 2. Average prediction error frequency for different types of audio streams (×10) 4 ) ; Table 2 shows the average prediction error distribution for six audio streams in the EBU-SQAM standard audio test set, which is widely used to evaluate the performance of data hiding schemes. As can be seen from Table 2, the prediction errors are mainly concentrated around 0, and the prediction error for 0 is significantly greater than for other values. Therefore, the content vector constructed using these prediction errors enables the RDH algorithm to achieve efficient data hiding while maintaining low distortion in the audio stream.

[0030] I. The process by which the sending end hides secret data in the audio stream, such as Figure 2 As shown: a. Divide the audio stream into point sets and intersection sets; using formula (11), through its 2 n The object audio point values ​​in the intersection set of the neighboring points are predicted to construct a sparse prediction error matrix. b. Divide the secret data into two equal parts, and according to step cd, hide the first part in the intersection set; c. Divide the cross set into two parts: the data hiding part E and the reserved part R; extract the least significant bit (LSB) of each audio point in the reserved part, and save the auxiliary data into the free LSB; d. Based on the location-free map scheme, the key data and least significant bit of the audio points are hidden in the reserved part and combined with part E; e. Using formula (12), predict the object audio points in the point set and their 2 n The values ​​of the nearest neighbor points are used to construct a sparse prediction error matrix for the invertible data hiding RDH; f. Use the same cross-set strategy as in step cd to hide the secret data in the point set; g. Construct a data-hidden audio stream using the fork set and point set data after hiding the information; II. The process of recovering the secret data and the original audio stream at the receiving end, and extracting the secret data, as follows: Figure 3 As shown: h. Divide the received audio stream into an intersecting set and a dotted set; i. Extract auxiliary information from the LSB of the retained part of the point set, and construct the prediction error matrix of the data hidden part E in the point set using formula (11); j. Based on the content of the extracted auxiliary data, extract the secret data and the LSB of the reserved part from the prediction error matrix of the data hidden part E in the point set based on the location-free map scheme; k. Recover the original amplitude of each audio point in the hidden data portion E and the point set preserved portion R; l. Using the same point set strategy as in steps j-k, extract the hidden secret data from the cross set and then restore the original value of each audio point; m. Reconstruct secret data using information extracted from point sets and intersection sets, and recover the original audio stream using the recovered audio points.

[0031] Experimental results and analysis of the present invention: In the experiments, the EBU-SQAM standard audio dataset was used to evaluate the performance of the localization-free audio reversible data hiding method based on additive vector stacking data hiding. This dataset contains 70 standard audio streams, each with 16-bit precision and a sampling frequency of 44.1 kHz, exhibiting a wide range of pitch, loudness, and timbre variations. To comprehensively validate the performance of the proposed scheme, six types of audio streams were selected, including wind instruments, percussion instruments, low-pitched music, instrumental music, orchestral music, and pop music. To ensure fairness in the comparison, the playback time for each audio stream was set to 10 seconds. Experimental results were evaluated using SNR-BPP curves for easy performance comparison.

[0032] Impact of the location-free mapping solution: In reversible data hiding, some carrier vectors may not be suitable for data hiding. Therefore, a location graph (LP) is used to mark these audio points to ensure complete data extraction and recovery of the original audio stream. Since the location graph needs to be hidden along with other data hiding coefficients (such as the expansion vector, data hiding strength value, etc.), this increases the amount of auxiliary data, significantly reducing the number of secret bits that can be hidden. This invention eliminates the need for a location graph in reversible data hiding by comparing the inner product between the expansion vector and the carrier vector with the 2-norm of the expansion vector, thereby correspondingly enhancing the data hiding capability of the proposed scheme. Figure 4 and Figure 5 The comparison results of the proposed location-map-free method are presented, showing that the proposed method achieves a higher signal-to-noise ratio (SNR) for the data-hidden audio stream compared to methods using location maps. Figure 4 As shown, at a load rate of 0.5 bpp, the average signal-to-noise ratio (SNR) of the data-hidden audio stream exceeds 61.3 dB; Figure 5As shown, the average signal-to-noise ratio (SNR) is 50.7 dB at a load rate of 1.0 bpp. These averages are 2.32 bpp and 1.71 bpp higher than when using a location map, respectively. Because the location map is eliminated, the space for auxiliary data hiding is significantly reduced, allowing more secret bits to be hidden in the object audio stream. This results in a gradual decrease in the SNR of the data-hidden audio stream as the data hiding capacity increases. Therefore, even at high data hiding capacities, the perceptual quality of the data-hidden audio stream remains good.

[0033] Comparison of this invention with other advanced solutions: To further evaluate the performance of the proposed scheme in data hiding, its superiority was demonstrated by comparison with some advanced audio RDH algorithms. Specifically, prediction was performed based on the audio points of the object and their past and future samples, and reversible data hiding was achieved through a PEE-based RDH scheme (Scheme 1). A pre-designed magic matrix was used to hide the secret bits into the audio stream, allowing the secret data to be hidden without being detected (Scheme 2). Currently, a PPVO framework has been proposed for reversible data hiding, significantly improving data hiding performance, especially for flat audio streams (Scheme 3). The above schemes demonstrate excellent performance in audio reversible data hiding. Therefore, these example schemes were used as benchmarks to evaluate the effectiveness of the invention through comparative analysis. The comparison results at data hiding capacities of 0.5 bpp and 1.0 bpp are shown in Tables 3 and 4, respectively.

[0034] Table 3. Performance comparison of the present invention with other advanced solutions at a load rate of 0.5 BPP. ; Table 4. Performance comparison of the present invention with other advanced solutions at a load rate of 1.0 BPP. ; As shown in Tables 3 and 4, this invention significantly outperforms other RDH algorithms in terms of signal-to-noise ratio (SNR) for data-hiding audio streams. On one hand, for smooth audio streams, such as wind instruments, bass songs, and pop music, this invention demonstrates superior SNR performance when the data hiding capacity is 0.5 bits per byte (bps) and 1.0 bits per byte (bps), respectively. Specifically, with a load of 0.5 bits per byte, the average SNR for bass songs reaches 83.73 dB, nearly 17 dB higher than Scheme 1. On the other hand, for audio streams with strong tones, such as percussion, instrumental, and orchestral instruments, the proposed scheme also outperforms other advanced schemes in data hiding performance.

[0035] Experimental results clearly demonstrate that this invention not only achieves high data hiding capability but also maintains the quality of the data-hidden audio stream. The reasons are as follows: Firstly, the RDH scheme based on additive vector superposition for data hiding requires changing at least two audio points when performing a single-bit hiding operation. Therefore, under low data hiding loads, this scheme results in greater audio distortion compared to other schemes. Secondly, due to the orthogonality of basis vectors, as the data hiding load increases, secret data can be repeatedly hidden into the object stream, causing most elements of the basis vectors to cancel each other out during multi-level data hiding. Therefore, under high data hiding capacity, the distortion of the audio stream is significantly reduced. In summary, the scheme proposed in this invention helps improve the perceptual quality of data-hidden audio streams, especially when the data hiding capacity is large.

[0036] This invention utilizes an additive vector superposition data hiding algorithm, which can completely recover the hidden secret data and the original audio. Due to the orthogonality of the basis vectors, the secret data can be hidden multiple times in the audio stream, achieving high data hiding capability. Simultaneously, during the multi-level data hiding process, most elements of the basis vectors cancel each other out, ensuring that the audio stream maintains high quality after data hiding, especially excelling in large-capacity data hiding. This invention explores the influence of basis vector length and data hiding strength to further evaluate the performance of the scheme. Furthermore, this invention proposes a location-graph-free scheme that significantly improves the hiding capability of the secret data by minimizing the size of the auxiliary data. Extensive experimental results demonstrate that this scheme outperforms other state-of-the-art RDH schemes in data hiding performance.

[0037] To enhance the data hiding capability of audio streams, this invention utilizes orthogonal extension vectors to hide secret data within the audio stream at the transmitting end, allowing the receiving end to reconstruct both the secret data and the original audio stream without loss. Furthermore, due to the orthogonality of the extension vectors, the secret data can be repeatedly hidden within the target audio stream, significantly improving its data hiding capability. During multi-level data hiding, most elements of the extension vectors cancel each other out, ensuring that the perceived quality of the audio stream remains high even with a large amount of data hidden. In addition, this invention designs a localization-free method to reduce the size of auxiliary information, further improving the capacity for hiding secret data.

[0038] Compared with the prior art, the present invention has the following advantages: 1. Based on the principle of additive vector superposition data hiding algorithm, specific orthogonal vectors are used to carry secret data and hide it in the target audio stream. Since these orthogonal vectors are linearly independent, the secret data can be repeatedly hidden in the audio stream, thereby enhancing the data hiding capability. Furthermore, due to the orthogonality of the basis vectors, during multi-level data hiding, most elements of the basis vectors cancel each other out, significantly reducing distortion in the data-hidden audio stream.

[0039] 2. By comparing the inner product of the basis vectors and the content vectors with the 2-norm of the basis vectors, this invention designs a method that does not require a localization graph. This method significantly reduces the auxiliary data required for reversible data hiding, thereby further improving the data hiding capability of audio streams.

[0040] 3. A high-precision adaptive target audio point numerical predictor was designed to estimate target audio points using their neighboring samples. Weights are adjusted based on the relationship between each sample and the target point, significantly improving prediction accuracy. Therefore, constructing a sparser prediction error matrix helps generate more suitable carrier vectors for data hiding.

[0041] The technical solution provided by this invention includes a data hiding algorithm based on additive vector superposition, which uses specific orthogonal basis vectors to carry secret data and hides it in the target audio stream; a localization-free map scheme is designed by comparing the inner product of the basis vector and the content vector with the 2-norm of the basis vector; and a predictor is designed to estimate the target audio point through its neighboring samples. This method reduces audio distortion caused by data hiding and improves the data hiding capability of the audio stream.

[0042] The various steps in the embodiments of the present invention can be performed by an electronic device. This electronic device includes, but is not limited to, tablet computers, portable PCs, and desktop computers.

[0043] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is running, it controls the electronic device containing the computer-readable storage medium to execute the above-described embodiment of the location-free audio reversible data hiding method based on additive vector superposition.

[0044] Figure 6 A schematic diagram of an electronic device provided in an embodiment of the present invention, such as... Figure 6 As shown, the electronic device 21 includes a processor 211, a memory 212, and a computer program 213 stored in the memory 212 and executable on the processor 211. When the computer program 213 is executed by the processor 211, it implements the localization-free audio reversible data hiding method based on additive vector superposition in the embodiment. To avoid repetition, it will not be described in detail here.

[0045] Electronic device 21 includes, but is not limited to, processor 211 and memory 212. Those skilled in the art will understand that... Figure 6 This is merely an example of electronic device 21 and does not constitute a limitation on electronic device 21. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0046] The processor 211 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0047] The memory 212 can be an internal storage unit of the electronic device 21, such as a hard disk or RAM of the electronic device 21. The memory 212 can also be an external storage device of the electronic device 21, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or FlashCard equipped on the electronic device 21. Furthermore, the memory 212 can include both internal and external storage units of the electronic device 21. The memory 212 is used to store computer programs and other programs and data required by network devices. The memory 212 can also be used to temporarily store data that has been output or will be output.

[0048] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0049] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for location-free, graph-based reversible audio data hiding based on additive vector superposition, characterized in that, The method includes: Step 1: Based on the additive vector superposition data hiding algorithm, a specific orthogonal extended basis vector is used to carry secret data and superimposed onto the object audio stream; Step 2: Based on Step 1, design a location-free graph solution by comparing the inner product of the basis vectors and the content vectors with the 2-norm of the basis vectors. Step 3: Based on Step 2, design an adaptive predictor to estimate the value of the target audio point with high accuracy through its neighboring samples, generate a sparse error prediction matrix, hide the data based on the sparse error prediction matrix, and improve the capacity of reversible data hiding.

2. The method according to claim 1, characterized in that, Step 1 includes: A. Data hiding: Assume the secret data sequence is , Secret position Convert to watermark data bits using equation (1) : (1) Based on equation (1), the secret data sequence is converted to: , ; Design basis vectors , Its elements take values ​​of 1 and -1, and its length is l The sum of all elements is zero; construct the set of basis vectors. Different basis vectors are orthogonal to each other; Assumption A It is a length of N The audio stream, from A Selecting adjacent samples to form a content vector Its expression is: , Let the length of the content vector also be... l The length of the content vector set is consistent with that of the basis vectors; the content vector set is represented as... ; According to the additive vector superposition data hiding principle, when the vectors satisfy the condition ,and At that time, the secret information bits are hidden in the content vector according to formula (2). middle; (2) ， (3) in, , , which indicates the first t A content vector that hides secret data; It is the secret bit to be hidden, with a value of 1 or -1; They are mutually orthogonal basis vectors; It is the number of hidden secret bits; on the other hand, when the vector satisfies the condition ,and At that time, according to formula (2), it will make or The basis vectors that hold true are superimposed on the content vector. Finally, the data hiding matrix is ​​obtained. ; When the amount of secret data is large, the orthogonality of the basis vectors can be used to hide the data sequence. The data is segmented and repeatedly superimposed onto the content vector to increase the data hiding capacity of the carrier audio stream, and its expression is given by formulas (4)-(6): (4) (5) (6) in, It is the set of selected orthogonal basis vectors; This represents the audio data matrix after multi-level data hiding; It is the number of times the data was hidden. It is the data hiding strength coefficient; B. Data Extraction: Assumption Matrix It can be seen that, composed of the data hidden content vectors in the received audio stream, , ,when Then, the hidden data can be extracted without loss according to formula (7): (7) No. t The basic principle of extracting the secret bits is shown in formula (8): (8) because and The values ​​of are all positive integers, therefore, when the condition is met... , At that time, the hidden data is extracted without loss according to formula (9): (9) When the condition is met , At that time, according to the formula The hidden data is extracted and discarded without loss, thus enabling the receiving end to recover the original secret information from the audio stream that carries the hidden data without loss. C. Original audio restoration: The original content vector is extracted from the data-hidden carrier audio stream, and the original audio stream can be reconstructed by subtracting the superimposed weighted basis vectors, as shown in formula (10): (10)。 3. The method according to claim 2, characterized in that, Step 2 includes: According to formula (9), when the condition is satisfied... and When, a bit is reversibly hidden into the carrier vector; when At this point, the inner product of the content vector and the basis vector is positive. Then, by additively superimposing the hidden data 1, we obtain... The receiving end can use this to determine that the vector has not been signed into secret data; when At this point, the inner product of the content vector and the basis vector is negative. Then, the hidden data -1 is added additively, resulting in... Based on this, the receiving end determines that the vector also does not hide secret data, thus achieving reversible information hiding of audio data without using a positioning map.

4. The method according to claim 1, characterized in that, Step 3 includes: According to the accuracy requirements of the hidden data, high-precision target point prediction can be achieved by adaptively adjusting the number of audio points involved in the prediction; large-capacity data hiding can be achieved by using the sparse error plane formed by the error of accurate prediction, as shown in formula (11): (11) in, , This is the prediction error value. It is the target point. yes The predicted value, , , and , , These are the nearest neighbor samples before and after the predicted target audio point. s It is the distance between the sample point and the target audio point; when s When we take 3, we get: (12) I. The process by which the sending end hides secret data in the audio stream: a. Divide the audio stream into point sets and intersection sets; using formula (11), through its 2 n The object audio point values ​​in the intersection set of the neighboring points are predicted to construct a sparse prediction error matrix. b. Divide the secret data into two equal parts, and according to step cd, hide the first part in the intersection set; c. Divide the cross set into two parts: the data hiding part E and the reserved part R; extract the least significant bit (LSB) of each audio point in the reserved part, and save the auxiliary data into the free LSB; d. Based on the location-free map scheme, the key data and least significant bit of the audio points are hidden in the reserved part and combined with part E; e. Using formula (12), predict the object audio points in the point set and their 2 n The values ​​of the nearest neighbor points are used to construct a sparse prediction error matrix for the invertible data hiding RDH; f. Use the same cross-set strategy as in step cd to hide the secret data in the point set; g. Construct a data-hidden audio stream using the fork set and point set data after hiding the information; II. The process of recovering the secret data and the original audio stream at the receiving end, and extracting the secret data: h. Divide the received audio stream into an intersecting set and a dotted set; i. Extract auxiliary information from the LSB of the retained part of the point set, and construct the prediction error matrix of the data hidden part E in the point set using formula (11); j. Based on the content of the extracted auxiliary data, extract the secret data and the LSB of the reserved part from the prediction error matrix of the data hidden part E in the point set based on the location-free map scheme; k. Recover the original amplitude of each audio point in the hidden data portion E and the point set preserved portion R; l. Using the same point set strategy as in steps j-k, extract the hidden secret data from the cross set and then restore the original value of each audio point; m. Reconstruct secret data using information extracted from point sets and intersection sets, and recover the original audio stream using the recovered audio points.

5. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the location-free audio reversible data hiding method based on additive vector superposition as described in any one of claims 1 to 4.

6. An electronic device, characterized in that, include: One or more processors; Memory; And one or more computer programs, wherein the one or more computer programs are stored in the memory, the one or more computer programs including instructions that, when executed by the device, cause the device to perform the location-free audio reversible data hiding method based on additive vector superposition as described in any one of claims 1 to 4.

Images